Mockitt AI MCP
Your AI assistant.
Your product workflow.
Generate product documents, prototypes and app drafts. Continue a task, inspect a prototype or export a canvas from an MCP-compatible AI client.
One connection, several ways to move a project forward
Create a first draft
Start a PRD, HTML prototype, React or Vue app draft, or image from a focused prompt.
Continue the same task
Find a recent generation task, check its result and request changes without starting over.
Read and export prototypes
Find prototypes you created, inspect their structure and export selected canvases as images.
Set up your connection
The integration uses Streamable HTTP and a personal-space token. Mockitt and Modao share the application, so the official server address and token header retain the Modao name.
1. Create your personal-space token
Open the AI application, sign in, then open the avatar menu and choose Token settings. Create a personal-space token and keep it private.
https://modao.cc/agent-py/ai/mcpSend the token in the modao-token header. The examples below use placeholders; replace them locally, never in a shared chat or repository.
2. Choose your AI client
Codex
Set the MODAO_TOKEN environment variable in the environment that starts Codex. Add the following to your user-level ~/.codex/config.toml, then restart the client.
[mcp_servers.mockitt]
url = "https://modao.cc/agent-py/ai/mcp"
env_http_headers = { "modao-token" = "MODAO_TOKEN" }Run codex mcp list or use /mcp to check the connection. The header value is read from the environment. See the official Codex MCP documentation.
Cursor
Open Cursor’s MCP configuration, or edit your user-level ~/.cursor/mcp.json. Merge this server into an existing mcpServers object rather than replacing other connections.
{
"mcpServers": {
"mockitt": {
"url": "https://modao.cc/agent-py/ai/mcp",
"headers": {
"modao-token": "YOUR_MODAO_TOKEN"
}
}
}
}Replace YOUR_MODAO_TOKEN with your personal-space token on your device. Save the file, refresh the MCP connection and inspect the available tools. Keep this file private. See Cursor’s MCP documentation.
Claude Code
Replace the token placeholder locally, then run this single-line command in your terminal to add a user-level HTTP connection.
claude mcp add --transport http --scope user mockitt "https://modao.cc/agent-py/ai/mcp" --header "modao-token: YOUR_MODAO_TOKEN"Open /mcp in Claude Code or run claude mcp list to check the connection. Take care with terminal history when using a token in a command. See Claude Code’s MCP documentation.
Other clients that support Streamable HTTP and custom headers can use the same server URL and modao-token header. Their setup screens may differ.
Verify the connection before you generate
3. Check your account
Ask your assistant: “Check my Mockitt account and credit status.” It should use get_account_status. Confirm the account is yours before granting access to a project.
4. Try a small task
For example: “Generate an HTML prototype for an equipment-loan request. Include item selection, date selection, confirmation and an unavailable-item state.” Review the result before expanding it.
If generation is still running, use get_task_result with the returned task ID. To refine that result, use continue_task; don’t start a duplicate generation.
Choose the output your next step needs
| Your task | Tools | What to know |
|---|---|---|
| Draft product requirements | generate_prd | Describe users, constraints, flows and acceptance needs. |
| Generate a prototype | generate_html, import_to_proto | HTML output can be imported into the prototype editor. |
| Explore an app implementation | generate_react, generate_vue | React and Vue projects cannot be imported directly with import_to_proto. |
| Continue existing work | list_recent_tasks, get_task_result, continue_task | Use the task ID to track or revise the same result. |
| Inspect a prototype | list_created_prototypes, get_proto_details, export_proto_to_png | Choose the project and canvas before requesting an export. |
Connection questions
Why does the server address use modao.cc?
Mockitt is the international site for the same application. Use the published endpoint exactly as shown, including the modao-token header name.
What if authentication fails?
Check that you created an application personal-space token, supplied the correct header and restarted the client after changing its configuration. For Codex, confirm the process can read MODAO_TOKEN.
How do I stop using the connection?
Disable or remove the server in your AI client. You can also delete the token in the application’s Token settings.
Where can I get help?
Contact support@mockitt.com with your client version and a redacted error message. Don’t include the token. The shared application’s MCP guide also documents its tools and additional clients.